Estimation of sib-pair IBD sharing and multipoint polymorphism information content by linear regression

A simple method of estimating IBD sharing (pi) of sibling pairs from multipoint genotype data based on linear regression is presented. The new method is an extension of that developed by Fulker, Cherny, and Cardon (1995) and involves a measure of sib-pair specific information, W, defined as the difference between the unconditional variance of pi and the conditional variance of pi given the marker genotype data. When markers are not fully informative, the W method provides estimates closer to those obtained from the exact hidden Markov method (HMM) than the original Fulker method. Using W, we also derive a generalisation for the polymorphism information content (PIC) for multiple markers. This multipoint polymorphism information content (MPIC) can be evaluated at any location along a marker map and is proportional to the noncentrality parameter for linkage. We illustrate MPIC by assessing the relative information content of single nucleotide polymorphism (SNP) and microsatellite markers.